Statistical Elucidations of The Seasonal Ambience of Physicochemical Characteristics of Coastal Waters Around South Andaman, India

The coastal areas of Port Blair in Andaman are threatened by severe water pollution due to the human settlements in these regions. The objective of this study was to determine the level of pollution in ten different anthropogenically dynamic coastal regions by assessing the seasonal variations of various physicochemical characteristics. The regions selected for this study were Burmanallah (BA), Carbyn's Cove (CC), Chattam (CH), Flat Bay (FB), HADDO Harbour (HH), Junglighat Bay (JB), Minne Bay (MB), Phoenix Bay (PB), Sisostris Bay (SB) and Wandur (WA) in South Andaman. The study was carried out from January 2018 to December 2018 to investigate seasonal variations in the physicochemical parameters such as pH, temperature, salinity, dissolved oxygen, nitrite, nitrate, ammonia, silicate, phosphate, and chlorophyll-α using multivariate statistical analysis. Statistical analyses suggest that the regions of JB and MB were highly polluted while, BA, CC, FB, and WA were the least polluted. Landuse and land cover analysis of the study area further facilitated and supported the multivariate statistical results.


Introduction
Coastal frontiers have been encroached on by human settlements since times immemorial due to the availability of rich resources and for logistical reasons as well. They are also the most vulnerable part of marine ecosystems (Jickells, 1998 (Shahidul and Tanaka, 2004) consequently resulting in the reduction of dissolved oxygen (Sanchez et al. 2007).
The coastal waters of Andaman and Nicobar Islands (ANI's) are experiencing higher rates of pollution owing to anthropogenically in uenced activities and surface land runoff contaminating the pristine marine ora and fauna Renjith et al. 2015). Nutrients from the domestic and urban sources are being drained into the coastal waters of South Andaman (Renjith et al. 2015). There are no sewage treatment plants in ANI's, so the contaminants are directly discharged into the adjacent sea through the nallahs. Nallahs are conduits of both rain runoff and sewage. South Andaman is the most populated and polluted maritime district of ANI's (Union Territory of India) as it houses the capital "Port Blair" and is the center for ANI's activities. This region being the capital is home to more than half of the island's population i.e. 1,44,418 with an urban population density of 4,402 individuals/sqkm (Census of India, 2011). Previous studies on the water quality parameters in ANI's are enumerated in Table 1.

Study Area
The study area between 11°27'00'' and 11°45'00'' N and 92°30'00'' and 92°46'47'' E, (Fig. 1), enjoys a tropical climate with an average annual rainfall of 3074.3 mm during 143 rainy days. The majority of rainfall (76.35%) is received during the Southwest Monsoon-SWM (May to September) followed by 22%, during a short spell of the Northeast Monsoon-NEM (October-December) and a dry spell during Pre-monsoon-PM, from January to April (1.64%). However, scanty rainfall is received by the study area during the dry spells of pre-monsoon (Fig. 2). The area chosen for this present study is around the capital of ANI's, is lined with sensitive and fragile wetland ecosystems such as seagrasses, mud ats, sandy beaches, and mangrove forests. It also has a tropical evergreen, semi-evergreen, and moist-deciduous forest with exuberant biodiversity and productivity.

Study design and sampling site description
The land use land cover (LULC) and sampling locations of the study are portrayed in the map (Fig. 1). The selected ten sampling sites were Chatham (CH), HADDO Harbour (HH), Phoenix Bay (PB), Sistoris Bay (SB), Junglighat Bay (JB), Minne Bay (MB), Flat Bay (FB), Wandur (WA), Carbyn's Cove (CC) and Burmanallah (BA). Out of the ten sites, WA, FB, and BA sites were less stressed by external factors. The remaining seven sampling locations are covered under densely populated Port Blair Municipal Council (PBMC) which releases sewage and other contaminants into the adjacent sea through various discharge outlets and drainage channels (Sarma et al. 2013). As per the Census of India (2011) urban and rural populations of south Andaman were estimated as 4,402 persons/sq.km and 37 persons/sq.km, respectively.
Contaminants like domestic sewage, apart from the wastes of the sawmill in Chatham (CH) drain into the adjacent sea. Haddo Harbour (HH) is the major harbor in ANI's and it shares space with the Indian Naval wharf. It is the busiest harbor as the mainland bound passenger vessels, cargo vessels, and oil tankers berth here. Apart from this, the surrounding area is densely populated. Huge quantities of pollutants are fed into the coastal waters due to the frequent movement of vessels and wastes from the dense human population. Phoenix Bay (PB) hosts inter-island transportation services and the surrounding hinterland is densely populated thus emptying its domestic sewage into the Bay. Sistoris Bay (SB) receives large amounts of anthropogenic wastes and sewage discharges. Junglighat Bay (JB) is a major sh landing center housing a dense shing community thus discharging domestic sewage, sh trash in the coastal waters of the Bay. Minne Bay (MB) has the waste dump yard of Indian Naval residential quarters.
Natural rivulets like Dhanikhari nallah, Mittakhari nallah and Ograbranj nallah drain into station Flat Bay (FB). Carbyn's Cove (CC) a pocket sandy beach is the most visited destination by both regional and international tourists. Another similar tourist destination is Wandur (WA) whose coastal waters are not as much stressed by anthropogenic activity unlike Carby's cove and also Wandur nallah drains station WA. The coastal waters of Burmanallah (BA) are polluted by the rural population in its local proximity.

Sample collection, preparation and analysis
Mid-day surface seawater samples were collected from the ten sites every month for a span of a year (January 2018 to December 2018). In each site, triplicate samples were collected for 12 months in a sterile polyethylene bottle. Parameters like pH, temperature, and salinity were recorded as in-situ measurements. The pH meter (OAKTON) with 0.000 accuracy was used, it had been calibrated with pH buffers (4 or 10). Standard Celsius Thermometer with 0.1°C accuracy was used for measuring temperature and a hand-held Refractometer (ATAGO) was used for measuring salinity.
In total three liters of water samples were collected for nutrient analysis, dissolved oxygen, and chlorophyll-a. Samples were collected separately, xed by Winkler's reagent and acetone 90% for DO assessment Winkler (1888) and the standard protocol was adhered for measuring chlorophyll-a (Parsons et al. 1992). The labeled water samples were refrigerated for further analysis. Millipore ltering system was used to lter the water samples which were later analyzed for nitrite (NO 2 -N), nitrate (NO 3 -N), phosphate (PO 4 -P), silicate (SiO 3 ), and ammonia (NH 3 -N) by adhering to standard analytical procedures as illustrated by (APHA, 1992; Clesceri et al. 1998; Grasshoff et al. 1999). The titrimetric method (Winkler method) was adopted for measuring the dissolved oxygen as modi ed by Strickland and Parsons (1972) and 90% acetone method, for measuring chlorophyll-a spectrophotometrically (Strickland and Parsons, 1972). The results of nitrite, nitrate, silicate, phosphate, and ammonia were expressed in µ mol/l. whereas DO and chlorophyll-a, were expressed in mg/l and mg/ m 3 respectively.  (Huete, 1988), and NDBI ( Zha et al. 2003) were employed for analyzing the land use and land cover. These ve indices models were developed in the model maker option of the ERDAS (2011) software. The outputs thus obtained were displayed in different combinations of red, blue, and green planes for differentiating and extracting LULC features through the generation of FCC's. Visually the FCC of NDWI, NDVI, and NDBI on red, green, and blue planes allows the discrimination of forest, built-up area, and wetlands. An FCC of NDWI, NDVI, and SAVI on red, green, and blue planes allows the decoding of inundated areas (tsunami-created wetland) and mangrove from the not-wetland areas. NDWI, NDTI, and NDVI on red, green, and blue planes allow nding wetland categories such as mangrove and creek easily.

Land Use and land Cover Analysis
LULC features deciphered from these techniques were also vectorized using Arcmap (10.5) software (Fig. 1).

Statistical analyses
Descriptive statistics such as one-way Analysis of Variance (ANOVA), Pearson's correlation matrix, Principal Component Analysis (PCA), and cluster analysis were used to assess the seasonal quality of physicochemical parameters in the ten sites using IBM SPSS-23 software. In general, One-way ANOVA is used to test interaction effects on multivariate data, using (approximate) permutation tests to avoid grossly unrealistic normality assumptions (Clarke et al. 2006). Statistical approaches like ANOVA were used by many researchers to ascertain signi cant differences between months and stations for physicochemical parameters (Jha et al. 2014, 2015a and b; Rajendran et al. 2018). PCA is a common but one of the most powerful techniques applied for minimizing the dimensionality of large data sets without redundancy of information. Also, the reduction of large datasets were accomplished by transforming the data set into a new set of variables called the principal components (PCs), which are orthogonal (non-correlated) and are arranged in decreasing order of importance. Mathematically, the PCs were computed from covariance or other cross-product matrices, which describes the dispersion of the multiple measured parameters to obtain eigenvalues and eigenvectors. The cluster analysis (Ward, 1963;Herion and Herion, 1995; Keenan and James, 2016) was used to detect the seasonal multivariate similarities/dissimilarities for physicochemical parameters (Farmaki et al. 2012). To assess the correlations between the levels of variables, correlation analyses were performed. Pearson correlation coe cients were calculated for a better understanding of the relationship between studied variables.

Physicochemical parameters
The minimum, maximum, mean, and standard deviation values of the measured physicochemical parameters from the chosen ten sampling locations are presented in Table 2  Biological processes include bacterial decomposition of organic matter in bottom waters, which could either be natural or induced by eutrophication (Kramer and Stein, 2003). The physical process that decreases the oxygen in coastal waters is vertical strati cation, which could be due to multiple reasons, including low tidal energy, large freshwater inputs, deep channels, and the presence of structures impeding circulation (Nixon, 1988  The concentration of phosphate was recorded high during NEM season and the least PM season (NEM > SWM > PM). JB station recorded the highest concentration of phosphate 0.14 ± 0.16 µ mol/l, 0.27 ± 0.19 µmol/l and 0.52 ± 0.07 µ mol/l when compared to other stations during PM, SWM, and NEM seasons respectively. The phosphate concentration of coastal waters is said to be linked to fertilizers and domestic wastewaters (Montaño and Robadue, 1995). Nutrients like nitrates and phosphates are brought into the coastal waters through river ow, agricultural and aquaculture runoff, industrial and household waste (Casali et al, 2007).

Statistical analysis
Cluster analysis, Pearson correlation, one-way ANOVA, and Principal Component Analysis were the four statistical analyses employed to address the seasonal ambiance of the physicochemical parameters in the focus area.

Cluster Analysis
Ward's cluster analysis was implemented to decipher the similarity and dissimilarity between stations and seasons ( Fig. 4a,  Cluster analysis of physicochemical parameters and their seasonal ambiance was applied to understand the similarity and dissimilarity among them (Fig. 5a, b, and c). Similar to the cluster analysis of stations and seasons, physicochemical parameters and seasons (PM, SWM, and NEM) also exhibit three distinct clusters among the parameters. Cluster 1 comprised of salinity and temperature in all three seasons suggesting that they are directly proportional to each other, i.e. if the temperature increases salinity increases and vice versa. Cluster 2 encompasses pH, silicate, and dissolved oxygen during NEM and PM season, while SWM season has only two parameters pH and silicate. The presence of silicate indicates that the focus area is devoid of landlocked watershed, meaning all the eroded materials are emptied into the adjacent sea as surface runoff (Ganeshamurthy et  3) In ux of silicate in the coastal water is due to tropical rains naturally ). The parameters of cluster 1 viz., temperature and salinity have a direct bearing on pH, and these three parameters together in uence the dissolved oxygen. Cluster 3 comprises nutrients like nitrate, nitrite, ammonia, and phosphate.
A comparison of both the cluster analysis viz., stations and seasons, physicochemical parameters, and seasons clearly emphasize the objective of the present study.

Pearson correlation
Correlations between ten water quality parameters were deliberated using Pearson's correlation, and the correlation analysis was used to describe the degree of relative correlation between the parameters. The correlation matrix describing the interrelationship between variables is presented in Tables 3a, 3b, and 3c for NEM, SWM, and PM seasons respectively. A high correlation coe cient (near + 1 or -1) means a good relationship, and there exists no relationship between two variables if the value is near zero at a signi cance level of < 0.05. To be precise, the parameters showing r > 0.75, 0.5 < r < 0.7 and r < 0.5 are considered as strongly, moderately and weakly correlated respectively (Gokul et al. 2018). The results for 10 water quality parameters in NEM (Table 3a) showed that a strong positive correlation existed between pH-nitrite (r = 0.  ). However, there was a discrepancy in correlation among parameters was observed seasonally. A high degree of strong correlation among the various parameter was observed during the PM season. Similarly, a high degree of moderate correlation was found among various parameters during NEM season.

One way ANOVA
The results of one-way ANOVA (Table 4)  ). Consistency in the physical parameters such as salinity and temperature (p > 0.01) was observed seasonally. Station-wise variation in DO (p < 0.013) suggests a differential in ux of nutrients, and freshwater due to anthropogenic intervention.

Land use land cover classi cation
Land use and land cover map (Fig. 1) of the area under investigation classi es the surrounding areas of the sampling stations.
Dense human settlements (urban) were around the stations via., HH, PB, SB, CH, JB, MB. The monsoonal stormwater and sewage are discharged into the adjacent marine environments through nallahs. An increase in population and related activities are the major source of pollution in the Bay regions. On the contrary, the pollution around rural areas surrounding the sampling stations FB, CC, WA, and BA are from the sources like small-scale farming and agriculture. It is also inferred from the land use and cover classi cation that the study area has a rolling topography and the domestic contaminants are directed to the nearby seas as run-off.
Thus the LULC map indicates the various sources of pollution of this region (Fig. 1).

Conclusion
The present investigation focused on the seasonal variation of physio-chemical parameters in ten stations of South Andaman Island for a year during 2018. The multivariate statistical analyses such as Cluster analysis, Principal component analysis, Pearson correlation, and one-way ANOVA suggest that the coastal waters of the stations Junglighat Bay (JB) and Minne Bay (MB) are polluted due to anthropogenic activities. Hence it is strongly recommended that these stations have to be monitored regularly and necessary initiatives should be taken by the Andaman administration to create awareness among the public regarding the importance of the health of these coastal waters. The present research also, strongly recommends proper waste treatment practices in Port Blair Municipal Council (PBMC), South Andaman. Further studies need to be carried out focusing on the fecal coliform content and heavy metal concentration of the discharge waters and the sites where the magnitude of the anthropogenic interference is high especially in JB and MB stations. This would provide a comprehensive view of the overall pollution levels of these waters.

Declarations
Ethics approval and consent to participate Neither human/tissue/body parts nor animal were used for the study: Not Applicable.

Consent for publication
Not Applicable.

Availability of data and materials
The data collected from the eld and analysed are presented in tables and gures.

Competing interests
The corresonding author and the co-authors do not have any con ict of interest.

Funding
The corresonding author and the co-authors declare that no funds were raised from national or international agencies for carrying out this research.  Tables   Due to technical limitations, table 1 & 2 is only available as a download in the Supplemental Files section.  Figure 1 The sampling locations and land use land cover of the study area Note: The designations employed and the presentation of the material on this map do not imply the expression of any opinion whatsoever on the part of Research Square concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. This map has been provided by the authors.   Table1waterqualityresearch.xls Table2AREASEASONWISENUTRIENTDATA.xls